Stereochemistry of the. alpha.-hydroxycarboxylic acids and related

Apr 1, 1977 - Jenny P. Glusker. Accounts of Chemical Research 1980 13 (10), 345-352. Abstract | PDF | PDF w/ Links. Article Options. PDF (1124 KB)...
0 downloads 0 Views 1MB Size
241 3

Stereochemistry of the a-Hydroxycarboxylic Acids and Related Systems Marshall D. Newton* and George A. Jeffrey Contributionfrom the Chemistry Department, Brookhaven National Laboratory, Upton, New York 1 1973. Received August 13, 1976

Abstract: The conformational energetics of three a-hydroxycarbonyl systems (glycolic acid, the glycolate anion, and glycolaldehyde) have been investigated using a b initio molecular orbital theory at the 4-31G level. The major aim of the study was to ascertain the role of intramolecular factors in the planarity or near-planarity of the carbon-oxygen framework observed in the cu-hydroxycarboxyl moiety of a large number of crystalline a-hydroxy acids and carboxylates, irrespective of the conformation of the a-hydroxy group. The calculations for the isolated species show that the equilibrium conformations correspond to internally hydrogen-bonded structures. Departures from equilibrium obtained by rotating the a-hydroxyl group are calculated to cost appreciably less energy than distortions which destroy the planarity of the carbon oxygen framework, a result consistent with the data for the molecules in the crystalline state, where stronger intermolecular hydrogen bonding takes precedence over internal hydrogen bonding. The conformational energetics of the isolated species are discussed in detail, and the nature of the internal hydrogen bonding is analyzed in terms of energy and charge density criteria and by comparison with related intermolecular hydrogen-bonded systems.

I. Introduction Since the earliest three-dimensional x-ray crystal structure determinations of the tartaric acids and tartrates, it was recognized that the three oxygen and two carbon atoms of the a-hydroxycarboxylic acid or carboxylate moiety form a characteristically planar or nearly planar group.' During the subsequent 20 years, more than 20 crystal structures have been reported containing molecules or ions of type I or 11, respec-

$ 1 = + 2 = +3 =

I

0"

+I =

$I2=

0"

II

tively, all of which show this conformational feature in varying degrees. They include the prototype glycolic acid (R, R' = H),2 the tartaric acids,3 tartronic acid,4 citric acid,5 and numerous crystal structures of glycol ate^,^ tartrates,' gluconate^,^ and

citrate^.^ These crystal structures have a variety of packing and intermolecular hydrogen bonding patterns, and it seems unlikely that they are all such as to place the same conformational constraint on the a-hydroxycarboxylic acid or carboxylate group. It is appropriate therefore to seek an explanation in terms of the intrinsic properties of the molecules or ioqs themselves. Unfortunately no structural data are available for the isolated species 1 or 11, but gas-phase microwavelo and infrared'' data have been obtained for the related system 111, the prototype carbohydrate, glycolaldehyde. Again a planar framework is found, with 41 = 180' and 42 = 0' as shown in 111 (the analogous conformation of glycolic acid is displayed as structure Ia)." This suggests that intramolecular hydrogen

bondingj3 may be a primary factor in the stabilization of the planar conformation in the isolated species. The theoretical calculations available for the various conformations of I11 support this vie\y.I4 On the other hand, inspection of the crystal structural data shows that such intramolecular hydrogen bonding is rarely observed. The values of 141I span a wide range from -0 to -1 7 3 O , as determined in most cases by the intermolecular hydrogen bonding. Yet the values of 42 are all close to Oo, and bear no obvious relationship to those of 41. This indicates that there may be other intrinsic molecular properties that restrict rotation about the C-C bond, such as hyperconjugative effects and interactions between electron lone pairs or bond dipoles. The purpose of this paper is to analyze the known conformational data for the species I, 11, and 111 and related systems and to compare them with the results of a b initio molecular orbital calculations, carried out at the 4-3 l G level," in the hope of discovering the role, if any, that intramolecular energetics plays in the observed persistence of the planar conformation in the crystalline state. In the course of this analysis we shall pay particular attention to the characterization of intramolecular hydrogen bonding. 11. General Background Before examining the experimental data and theoretical results for I, 11, and 111 in greater detail, we broaden the scope of this study by noting that these molecules can be viewed as specific examples of a more general class of carbonyl systems, IV, for which a great amount of conformational data is avail-

m

..............0 d l = 180°,+2=00

m

able, both experimental and theoretical. The discussion is confined to various acetyl derivatives (R, R' = H), with Y taken as H, CH3, or an electronegative group (OH or a halogen). W e first deal with the simplest case, acetaldehyde (X, Newton, Jeffrey

/

Stereochemistry of a-Hydroxycarboxylic Acids

2414 Y = H), which is characterized by an equilibrium conformation with one of the methyl C H bonds trans staggered relative to the aldehyde C H bond (42 = Oo).18 The alternative statement of this situation, namely that a methyl C H bond prefers to eclipse the CO double bond, appears to be a t odds with expectations based on steric interactions, although perhaps not so if the CO double bond is interpreted as two bent single bonds, which would then bear a more acceptable gauche relationship to the methyl C H bond. At any rate simple arguments from perturbation theory19 based on the a and a* CO bond orbitals and the corresponding ?r-type orbitals formed from the methyl C H bond orbitals can rationalize the observed equilibrium angle ( $ 2 = 0"). and experimentI8 both indicate a threefold rotational barrier of -I kcal/mol. The situation is formally the same when the substituent Y is changed to an electronegative group (e.g., NH2 or O H ) , although the rotational barrier is appreciably smaller (0.3-0.4 k~al/mol).'~ Perturbing the methyl group of acetaldehyde with an electronegative group X can have several conformational consequences. The hyperconjugative interactions between the methyl and carbonyl groups will clearly be affected, and no definitive prediction of the preferred conformation based on these interactions above seems possible. Other effects may be expected to dominate; e.g., nonbonded repulsion between the electrons of X and those of 0 and Y, or dipole-dipole interactions among the CX, CO, and C Y bonds, which would be expected to favor 42 = 180" if Y were H or CH3. Indeed, gas-phase infrared2' and microwave22 data for a-halo aldehydes and acetones (X = F, CI, Br; Y = H or CH3) indicate a large equilibrium value for 42 (estimated at 2 150' for X = CI), although another isomer with 42 -0" was also detected. This rotational isomerism is observed in the condensed phase (liquid, solution, solid) as well. Molecular orbital calculations for fluoroacetaldehyde (X = F, Y = H)lhc indicate an energy separation of -4 kcal/mol between the 42 = 0" and 42 = 180" isomers. I n contrast to the ketones and aldehydes, the a-haloacetyl halides ( X , Y = F, CI, Br) are found to prefer the 42 = 0" isomer by 1-2 kcal/mol over a second isomer with 42 2 1 SO", reflecting a delicate balance between the orientation of the CX, CY, and C O bond^.'^.^^ The same qualitative ordering is observed in the liquid. In the related molecule 2-fluoroacetic acid (Y = O H ) the energy separation of the two isomers is somewhat smaller (0.6 kcal/mol), possibly due to internal hydrogen bonding in the 42 = 180" isomer (with 43 = 180"; see structure I).25 I f the substituent X is changed from a halogen atom to an electronegative group like O H or NH2, then an additional degree of the conformational freedom is introduced, and for appropriate values of 91 and $ 2 the group can at least formally serve as a proton donor in an intramolecular hydrogen bond, with either the carbonyl oxygen or an electronegative group Y acting as the proton acceptor. A third H-bonded conformation would have a carboxylic group (Y = O H ) serving as proton donor to the electronegative cy substituent ($1 = O", 4 2 = 1 SO", 43 = 1 SO"), as discussed above for the case of 2-fluoroacetic acid. It is interesting to compare the acetyl halides with the structurally similar species I (both have electronegative X and Y groups), since the former systems demonstrate that preference for 42 = 0 can occur in the absence of any possibility of intramolecular hydrogen bonding, both in the gas and liquid phase. The halogen derivatives also suggest that rotational isomerism might be observable for I and I I I . With molecules of the latter type having been placed in the general context of species IV, we now turn to a detailed discussion of their properties. Further interest is provided by the relationship of 1 and I I I to their isoelectronic counterparts obtained from replacing O H by NH2 in the cy position. Calculations for these amino Journal of the American Chemical Society

/

993

derivatives have been carried out by Vishveshwara and Pople.26 111. Structural Details Values of the conformational angles 41and 42 and the Hbond distanced (see structure la) from the crystal structures are given in Table 1. The standard deviations on 4 2 are of the order of 0.5". Those on 41are up to ten times greater unless a neutron diffraction study has been made, in which case they are comparable. Values of 43 for the undissociated acids are in almost all cases very close to 0°.27 In the majority of the structures 1421 is less than IO", making the carbon and oxygen atoms close to coplanar, with the hydroxyl and carbonyl oxygen cis. A 142) value of 10" corresponds to a displacement of the hydroxyl oxygen from the carboxylate plane by -0.25 A. The sole example of rotational isomerism (cf. Section 11) is provided by the monoclinic form of meso-tartaric acid, where in one half of the molecule the hydroxyl and carbonyl oxygens are trans (i.e., 4 2 --180°).3b Aside from this exceptional case the largest observed value of 1421 is 19", in tartronic acid.4 The clustering of 42 values near 0" is consistent with the known conformations of the related halogen derivatives discussed in the previous section. However, in the case of 1 , I I , and I l l additional stabilization of the 42 = 0" conformation might be obtained from internal H bonding (with 41 = 180" and d -2.0 A), even though such a conformation would involve eclipsing of the O"H and C C bonds. For example, structure I I I is observed for glycolaldehyde,10 and the previously mentioned calculations indicate that the equilibrium conformation ($1 = 180°, $2 = 0 " ) lies 7.2 kcal/mol below the alternative open, staggered structure with $ 1 = 0" and 4 2 = O.".I4 A short-range hydrogen-bond interaction undoubtedly makes some contribution to this energy difference, but one must assess the importance of other factors such as barriers to rotation, and dipole-dipole and other nonbonded interactions. In spite of our expectations with regard to intramolecular H-bonded structures we find in fact only two unambiguous examples of such a conformation from the crystal data for molecules of the type I and I I . They occur in sodium D-tartrate dihydrate, where 41 = 173", 42 = So, and d = 2.04 A,7aand in potassium gluconate monohydrate (form A), where 41 = 146", 4 2 = 6", and d = 2.12 A.s The latter is a neutron diffraction study, which shows that this hydrogen bond is bifurcated, half intra- and half intermolecular, with an intermolecular 0.-H distance of 1.98 A. Less definitive examples are from the x-ray studies of tartronic acid,4 and the anhydrous and monohydrate citric acid structure^,^ where the intra- and intermolecular H - 0 distances are 2.23 and 3.02, 2.29 and 2.1 5, and 2.29 and 2.03 A, respectively. The persistence of 42 -0" even when 41is close to 0" is noteworthy in that this latter orientation of the O"H bond directs the hydroxyl lone pairs toward the carbonyl oxygen in a situation where the two oxygens are separated by only 2.7 A and thus appears to be slightly within the normal nonbonded (van der Waals) contact.28 Due to the occurrence of such close contacts, one might have expected a correlation between $1 and 42. The data in Table I show clearly that this is not the case. The angle 4 2 is close to O", irrespective of 41.Close inspection of the crystal structures shows that with the exception of the cases of internal H bonding noted above, the observed 4'values are determined primarily by the requirements of intermolecular hydrogenbond formation, either to adjacent like molecules or to water molecules. It thus appears that, as in the carbohydrate^,^^ intermolecular packing is such as to provide more favorable geometry for intermolecular hydrogen-bond formation than can be obtained internally for I or I I without considerable bending of valence angles from tetrahedral and trigonal. Given this phenomenological indication that intermolecular

/ April 13, 1977

241 5 Table 1. Crystallographic Data" Conformational angles, deg

41

Molecule

$2

63

d,Ab

Ref

Key to Figure 2

(A) a-Hydroxycarboxylic Acids

Glycolic acid' anhydrous ( N ) d Tartaric acid' D-monohydrate ( N ) meso-monohydra te (triclinic) meso-monohydrate ( monoclinic) meso-anhydrous (triclinic) Acid group in monohydrogen tartrates D-anhydrous ( N H 4 + monoanion) me.w-an hydrous' ( K + monoanion) Tarlronic acidn anhydrous Citric acid" anhydrous monohydrate

100 100

-6 2

-3 -3

2.95 2.90

2

a b

25 -109 -3 12 -6 26 -19 69

-2 -6 -4 -6 9 I70 -8 7

-2 7 -6 -I -6 IO 7 -39

3.58 2.8 1 3.34 3.29 3.47 3.45 3.30 3.34

3a

C

-6 1

7

-4

3.37

7b

k

I

-3

2.74

7d

1

16

-13 9

2.23 2.93

4

m

0 12

2.29 2.29

5a 5b

7

2.86

6

I73 123 - 1 I7

8 15 -5

2.04 2.86 2.7 1

7a

-146 130 - I 18

-10 4 2

2.48 2.78 2.64

7c

146 114

6 3

2.12 2.64

8

3.44

9a

3.49

9b

-105 I47 -9 1

-19

12 143 2 I60 (B) a-Hydroxycarboxylates

Glycolate a n hydrous (Lit monoanion Tartrate tl-d i h yd ra te ( 2 N a + dianion) wanhydrous ( N H4+ monoanion) meso-dihydrate (K+ dianion) meso-an hydrous ( K + monoanion)/ Gluconate' D-monohydrate ( N ) Phase A ( K t monoanion)] Phase B Citrate anhydrous ( N a t monoanion) monohydrate (Li+, NH4+ dianion)

-99

-14

-17

0

4

d e f g h

3b 3b 3b

1

j

n o

P

7b

7d

('Structures included are those for which reliable values of the parameters @ I , $2, 43, and d are available. In most cases these quantities were not explicitly reported and had to be calculated from the unit-cell dimensions and the atomic coordinates. All the data are from x-ray single-crystal structure determinations except where denoted by ( N ) for a neutron diffraction analysis; D refers to the absolute configuration. Defined in structure I. C H 2 0 H - C O O H : see structure I . Two independent molecules in unit cell. e (CH20H-COOH)2. Two sets of values for $1, $2, $3, and d a r e also available for the dimethyl ester of tartaric acid; I 18, -I, -7O, 2.64 A, and I36,0, -Io. 2.52 8, (J. Kroon and J . A . Kanters, Acfa Crystallogr., Sect. B, 29, I278 (1973). /Although two independent molecules a r e found in the unit cell, two of the carboxyl groups are coupled across a center of symmetry in a strong ionic intermolecular hydrogen bond of the type (-O-.H-.O-)-. The @ values associated with these carboxyl groups have not been included in the table since the intramolecular conformations might be expected to be dominated by the strong intermolecular interaction. Even in this case, however, the $2 values are close to 0' as usual, although one of the two 43 values is close to 180' in contrast to the other crystal data, as noted in ref 27. g H O O C - C H O H - C O O H (a-hydroxymalonic acid). HOOC-COH( C H I C O O H ) ~ '. Anion of HOOC-(CHOH)4-CH20H. J From two different crystal structures.

*

forces play an important role in the crystal conformations of I and I I , it still r e m a i n s to determine whether the observed distribution of conformational angles reflects or is consistent with features of the intramolecular conformational energy surface. While calculations are reported below for all three species, I, 11, and Ill, most attention will be given to I and 111, since the stronger intermolecular forces associated with the ionic species I1 presumably make its intramolecular energetics less relevant to its conformations in crystalline environments.

IV. Methods and Results

Standard, closed-shell, self-consistent molecular orbital theory with the 4-31G extended Gaussian basis set was used.I5 The molecular structure of glycolic acid ( I ; R, R' = H ) was based on the accurate neutron diffraction study of the crystal.2 A fixed set of bond lengths and angles was adopted for all conformations and is illustrated in Figure 1 for 41,42,and 43 = 0'. The conformational energy was calculated as a function of the angles $1 and $2 at intervals of 60' and is presented in

Newton, Jeffrey

/

Stereochemistry of a-Hydroxycarboxylic Acids

2416 0

Table 11. Relative Conformational Energies of Glycolic Acid and Glycolaldehyde

1.093A H

Conformational angles, dega

41

1.463

1

Figure 1. Bond lengths and bond angles employed in the glycolic acid calculations. A local symmetry plane (C, point group) was assumed for each of the carbon atoms. The figure corresponds to structure la ( $ , , $ I ,

97 = O O ) .

$I2

0

0 60 12p t 80 0 60 120 180 -120 -60 0 60

60

120

I20 180 -120 -60

60'

0

180

60 I20 180

1200

P

180'

- I200

Energy, kcal/molb Glycolic acid' Glycolaldehyded 6.0 5.7 2.0 0.0 8.2 7.2 4.3 8.2 7.7 8.6

I .o

7.3 6.5 1.9 3.8 6.1 7.0 1.2 4.0 2.7

6.4 6.4 2.4 0.0 8.1 7.3 4.2 7.6 8.0 9.3 4.1 4.1 3.7 1.3 5.3 5.4 2.3 2.8 2.1 5.4

Angles are defined in Section I1 (see structure I ) . All data are based on 43 = 0'. Relative to the minimum energy conformation (41 = 180°, $I2 = 0'). Total energy for $ 1 = 180, $12 = 0' is -302.20134 au. Total energy of 41 = 180°, $2 = 0' is -227.42259 au.

-60°

Table 111. Relative Conformational Energies of the Glycolate Anion no

--mo

-120

-60

o

60

i20

180

Conformational angles, dega 41 42

42

Figure 2. Conformational potential energy map for glycolic acid (based on 43 = 0'). Contour labels refer to energy (kcal/mol) ielative to 91 = 180'. $2 = 0". Roman numerals designate conformational structures referred to in the text. They are not included on the left-hand side of the figure, which is related to the right-hand side by inversion through the The lower case letters identify crystal data center ( @ I = 180, $2 = points listed i n Table IA (see the last column). Note that the letter "a" is near structure VI and should not be confused with structure Ia.

Table I 1 and also as a potential energy contour map in Figure 2, with contours separated by 1 kcal/mol and ranging from 1 to 8 kcal/mol. The energies are all relative to ;he minimum at $ 1 = 1 80°, $2 = 0'. The letters in Figure 2 refer to the crystal data points for the neutral acids listed in Table I (see last column) and the Roman numerals correspond to the conformations discussed in the text. The third conformational angle (43) was kept at 0", in keeping with the very small values generally observed in the crystal structure^.^^ However, conformations with 43 = 180" are considered briefly below. The calculations for the glycolate anion employed the glycolic acid framework geometry, except for the carboxylate group, which was given local CzLsymmetry with r c o = 1.26 .& and +OCO = 125", based on crystal structure data.6 Calculated results are given in Table 111. The geometry for I11 was the same as that for I (Figure I ) , except for the replacement of OH by H at the carbonyl group, with rCH = 1.10 A and +HCC = 118", similar to theanalogous values in acetaldehyde.18 This choice for I11 was thought to be somewhat preferable to alternatives, such as the standard molecular geometry30 which was used in the previous molecular orbital studies.I4 The main difference is the significant departures from 120" bond angles a t the carbonyl carbon atom. At any rate, comparisons show that the reiative conformational energies based on the geometry used here and on Journal of the American Chemical Society

0 120 180 0 120 180

0 90

Energy, kcal/molb 18.2 9.3 0.0 19.5 13.6 13.6

See footnote a , Table 11. The angle $3 is not defined for the anion. See footnote 6, Table I I . The total energy for $ 1 = 180°, 62 = 0' is -301.64808 au.

the standard geometry differ by 5 1 kcal/mol. The conformational energy of I l l with respect to $1 and $ 2 is given in Table I 1 and also in Figure 3 as contour levels relative to the minimum a t $1 = 180" and $ 2 = 0". Figures 4 and 5 contain various one-dimensional profiles comparing the conformational energetics of I and 111, with related curves for ethanol and acetaldehyde presented for comparison. Some data pertinent io hydrogen bonding in I and 111 are given In Table I V . After the calculations for 111had been completed we became aware of the recent microwave structurelob (a refinement of earlier workloa which had tentatively postulated structure 111), which is quite similar to the one employed here except for the geometry of the hydroxyl group. In spite of the anticipated weakness of the intramolecular H bond due to the strained geometry (iO"H-.O = 121°, sH-O=C = 84'); the microwave structure nevertheless suggests a rather strong interaction: i.e., a sharply reduced HO"C bond angle relative to methanol (101 vs. 107") and a large elongation of the OH bond length (1.05 vs. 0.96 A).31,32 The latter effect is surprising, since even in very strong ionic hydrogen-bonding situations the OH elongation is only -0.25 (e.g., the aquated hydronium

/ 99:8 / April 13, 1977

2417

n

I

I

1800

I

60"

1200

00

+I

Figure 4.One-dimensional conformational energy profiles for glycolic acid, glycolaldehyde, and ethanol: variation with respect to 4 1w#ith$ 2 fixed at

00.

-

7

I ( a ) 42 =Oo

GLYCOLALDEHYDE

180°

42

Conformational potential energy map for glScolaldehyde. Contours are labeled as in Figure 2. Figure 3.

and neutral H-bonded systems such as the water dimer exhibit very little lengthening (-0.01 Microwave structures for 2-amino- and 2-haloethanols have also given indications of unusual O H elongation, presumably due to H bonding.j6 Accordingly, we simultaneously varied the total energy of structure 111 (4, = 180°, 4 2 = 0") with respect to +HO"C and rOH. The calculations indicated no significant stretching of O H or reduction of the bond angle relative to the methanol values.37 These details of the OH-bond geometry clearly warrant further attention in the future, both theoretical and experimental. In previous studies, the 4-3 l G basis has been shown to give good quantitative account of OH-bond lengthening due to hydrogen bonding.34b

V. Discussion I n the following discussion we first point out the close correspondence between the crystal structure data and the a b initio conformational energy surfaces of glycolic acid and the glycolate anion.38The surface of glycolic acid is then examined further, with particular emphasis on the characterization of internal hydrogen bonding. Finally, comparisons are made between glycolic acid and related systems such as glycolaldehyde (111) and the nonhydroxylic species mentioned in Section 11. A. Analysis of the Crystal Structure Data. The energy calculations show (Figure 2) that the isolated glycolic acid molecule prefers a conformation with $ 1 = 180" and $2 = O", which is the optimum geometry for intramolecular hydrogen-bond formation. In the crystalline state, and probably in aqueous solutions also, intermolecular hydrogen bonding is energetically more favorable. This, however, requires a conformational change, depending on the molecular packing, to permit the hydroxyl hydrogen to be directed toward the hydrogen-bond acceptor atom on the adjacent molecule. The energy contour map, Figure 2, shows that in general such a conformational change can be achieved more easily by changing $1 than 42. A 90" change in $ 1 , for example, can be made a t the cost of somewhat less than 4 kcal/mol, whereas for $2 this would require more than 8 kcal/mol. The distribution of the experimental points along the $ 1 axis is therefore consistent with the theoretical result in that this is the direction in which we observe the greatest range of conformational change, while $2 remains relatively constant and close to 0". The clustering of $2 values near 0" reflects the fact that for any value of $1, the minimum internal energy always occurs for $2 = 0". The greatest intramolecular energy cost is seen to be

1

0'

--I---i--1 60"

1200

180"

$2

Figure 5.One-dimensional conformational energy profiles for glycolic acid, glycolaldehyde, and acetaldehyde: ( a ) variation with respect to 9,. with $1 set equal to 180Oor defined by the minimuni energy path: ( b ) variation with respect to $ 2 , with $ 1 fixed iit 0'.

associated with those experimental points whose 191I values are near O", but this energy (- 6 kcal/mol) can clearly be recovered by intermolecular H bonds. Consistent with this picture is the fact that the molecule whose crystal structure conformation lies the closest to the intramolecular energy minimum-Le., tartronic acid (vide supra)-is one whose w h y droxy group is not involved in any intermolecular hydrogen bonding. A comparison of the one-dimensional profiles for glycolic acid in Figures 4 and Sa gives further illustration of the relative sensitivities of the conformational energy with respect to 41 and $ 2 . The energy in the vicinity of structure la rises much more sharply with $2 even on the minimum energy path (i.e., based on the minimum energy $ 1 value for each $2). Analogous calculation^^^ for the isoelectronic species, glycine, reveal a much flatter energy as a function of 4: because the two protons on the donor NH2 group allow considerably greater flexibility in H-bond formation, including the possibility of multiple H bonding (Le., two bent N H - 0 bonds to the carbonyl oxygen). It is apparent that while the $ 1 profile for glycolic acid is monotonic (Figure 4), the 42 profile finally descends as 42 approaches 180" (Figure 5a), corresponding to the presence of a second local minimum (see Figure 2 and structure V ) . This

Newton, Jeffrey

H

H

P

/ Stereochemistry of a-Hydroxycarboxylic Acids

2418 Table IV. Population Analysis of tlie -0"H.

. .O-

Linkagea.6

Atomic population shifts (Aq)c

Overlap populations

H-bond energy, kcallmole

Species

A90"

Glycolic acid (Ia) Glycolaldehyde (111) Propane-1 ,3diolb (gauche reference structure) (all-trans reference structure)

0.034 0.034

(A) Intramolecular H Bond -0.025 0.036 0.032 -0.024 0.041 0.042

-0.012 -0.008

0.005 0.002

0.061 0.020

-0.061 -0.049

-0.052 -0.052

0.006 -0.005

(9.5) (0.04)

AqH

*qo

0.05 2 0.011

OPo.. .H

0.040 0.040

OPo..

AOPO"H~

(6.1) (6.4)

(B) Intermolecular H Bond -0.068

0.043

0.046

-0.030

-0.017

5.7

-0.056

0.011

0.050

-0.018

-0.015

8.1

a All population changes (Aq, AOP) and H-bond energies are defined relative t o nonhydrogen-bonded reference species: the reference for structures Ia and I11 is the = 0, e2= 0" conformation (structure I, for glycolic acid); the reference for each dimer is the associated pair of isolated monomers. bTwo different nonhydrogen-bonded reference structures have been considered for propane-l,3diol, as defined in ref 42a. CPositive Aq corresponds to an increase in atomic electron population. dPositive AOP corresponds t o an increase in tlie population of the O"H bond. eTlie intramolecular hydrogen-bond energies are placed in parentheses to emphasize their dependence on the selection of a reference conformation. fBased o n the same 0 . .O distance (2.73 A) as in structure Ia, but with a linear hydrogen bond ( H . . .O = 1.76 A). For comparison, a similar calculation based on the H...O distance of structure Ia (2.22 A), and hence with yo...o = 3.19 A , yields essentially the same bond energy (5.8 kcal/mol), hut with somewhat smaller populations and population shifts: 0.038, -0.030,0.020, 0.039, -0.006, and -0.01 2 , respectively.

minimum, like that represented by structure Ia, involves internal hydrogen bonding, which will be discussed in the following paragraphs. I n the crystal structures, neither of these intramolecular hydrogen-bonded conformations is populated, due primarily to the fact that more favorable intermolecular H bonds can generally be formed. Some insight into the skewing of the central basin in the 41 direction (Figure 2) is offered by considering, for example, the structures obtained from Ia by varying either $1 or 42 by 60°, VI and VII, respectively. Structure VI1 is calculated to be 6.2

m kcal/mol less stable than VI. The 0"-0 and distances are very similar for the two structures (2.73 A for VI and 2.96 A for VII), and the 0"-H distances (>2.6 A) indicate that these 60" distortions from Ia have eliminated any hydrogen bonding. Perhaps the most important factor contributing to the 6 kcal/mol difference is that the process Ia VI leads to relief of eclipsing between the O"H and CC bonds, while the process la VI1 creates additional eclipsing of single bonds (CO' and CH). The relative energies of structures la, VI, and VI1 could not be cqrrelated in terms of appreciable differences in the overlap population of the CC bond. Such differences might be expected if differential hyperconjugation were playing a n important role. The remarks made above with regard to the distribution of the $1 and 42 values for the undissociated a-hydroxy acids can also be applied to the a-hydroxycarboxylate anions (see Table 1B and Table 111). The fact that the values for the anion are appreciably closer to 180" than in the case of the neutral (8 of the 1 I $ 1 values are within -60" of 1 80") probably reflects the much sharper rise in energy with respect to 41 in the vicinity of 42 = 0, as revealed by the calculated results for the glycolate anion. B. Intramolecular Hydrogen Bonding. In spite of the general absence of internal hydrogen bonding in the crystal structure data already discussed, it seems clear that such bonding has an important bearing on the conformational behavior of mol-

-

-

Journal of the American Chemical Society

ecules like glycolic acid when isolated or in a nonhydrogenbonding environment, and we consider the matter in some detail. Geometrical Characterization. The geometry of the minimum energy conformation (structure Ia) suggests a rather distorted internal hydrogen bond: ro"...o = 2.73; rot' ..H = 2.22 A; sO"H-0 = 112; and -+H-.OC = 81". In an unstrained situation, the latter two angles would be expected to approach -1 80 and -120°, r e ~ p e c t i v e l y The . ~ ~ second local minimum at 41= 180°, 42 = 180" noted in Section VA (structure V) also possesses a bent hydrogen-bond geometry (TO...H = 1.96; r o...0 = 2.53 A) and lies 2.7 kcal/mol above la (an analogous separation of 2.5 kcal/mol is found for glycine26). The latter fact may appear surprising, since rot!...H is shorter than in Ia (1.96 A). However, within a carboxyl group the carbonyl oxygen is expected to be a better acceptor than the hydroxyl oxygen.40 In addition, the C O single bonds are eclipsed in structure V, and with the rigid rotor geometry adopted here the 0"-0 contact is quite close (2.53 A). Rotation of both $1 and 43 by 180" in structure V leads to another hydrogen-bonded interaction of the two hydroxyl groups (structure VIII), with the donor and acceptor roles of 0' and 0" interchanged. Structure VI11 lies -2.7 kcal/mol H F

Y\/

H\o,l/

I =+ =180" 2 3

Ho'' +=OQ,+ I

Trm: above structure V (or 5.4 kcal/mol above la). Although 0" is expected to be a better proton acceptor than 0' (see ref 40), the fact that V is more stable than Vi11 can be rationalized as follows. The eclipsing of the O"H and CC bonds destabilizes V relative to VI11 by -2.1 kcal/mol (from comparable studies of ethanolI4), but the cis conformation of the carboxyl group (H-0-C=O) stabilizes V by -6.3 kcal/mol (from comparable studies of formic acidI4), since VI11 contains the less favorable trans conformation. If it were not for internal hydrogen bonding, V should therefore be more stable than VI11 by' 4.2 kcal/mol. The net calculated value of 2.7 kcal/mol implies that the hydrogen bond in VI11 is more stable than that in V by -1.5 kcal/moL4'

/ 99:8 / April 13, I977

2419

Charge Distribution and Energetics. Charge distribution and energetics provide additional criteria for assessing the extent of hydrogen bonding. The reliability of the 4-31G basis for analyzing differences in hydrogen-bond strength has recently been documented in a study which systematically considered a variety of atomic orbital basis Typical weak intermolecular hydrogen bonds are characterized by a stabilization energy of a few kilocalories per mole relative to isolated monomers, a gain in the electronic population on the electronegative donor and acceptor atoms (oxygen, in the present case), and a decrease in the population of the donated proton.17 These criteria can also be applied to intramolecular H bonding, although relatively few detailed studies of such systems exist.42 The intramolecular H-bond energy cannot, of course, be based on isolated monomers, but may be defined in terms of an appropriate non H-bonded conformation of the molecule, thereby introducing a certain degree of arbitrariness. The process of converting the nonhydrogen-bonded conformation into the hydrogen-bonded one not only leads to short-range interactions between the donor atom proton and the acceptor atom lone pair, an important component in any hydrogen bond, but can also cause appreciable energy changes due to variation in dipole-dipole orientations and local barriers to internal rotation. The analysis of the internal hydrogen bonding in structures la and I l l is summarized in Table IV, which includes a comparison with an analogous intermolecular interaction between formic acid (acceptor) and methanol (donor), all geometrical parameters of the monomers having been taken from the corresponding glycolic acid values (Figure 1 ). This dimer is depicted in structure IX. Values of I80 and 120’ were assigned

H

Ix to a1 and ( ~ 2 respectively , (the system is planar except for two of the methyl protons), and the 0-0” distance was taken as 2.73 A, the same as in structure Ia. Further comparisons are provided in Table IV by the water dimer and propane-l,3-diol, two systems which have in the past stimulated great interest as examples of inter- and intramolecular hydrogen bonding, respectively.’ 3,17,42a The population analysis data for structures Ia and 111 exhibit the characteristics of hydrogen bonding noted above, but suggests a substantially weaker interaction than that observed in the intermolecular cases. In this connection, the nature of the donor OH bond merits some attention. A traditional probe of hydrogen bonding, both inter- and intramolecular, has been the red shift of the O H stretching frequency,I3 attributed to weakening of the donor O H bond, although in intramolecular cases the relationship between H-bond strength and real shift appears to be quite complicated36d (see also the theoretical discussion in ref 42b). It is therefore of interest to examine the donor O H overlap populations in Table IV. As opposed to the intermolecular cases, the populations for the hydrogen-bonded forms of glycolic acid and glycolaldehyde are slightly larger than those for the reference molecules. A similar absence of O H weakening is found for propane- 1,3-diol when the non-H-bonded reference structure has the same carbon-oxygen framework conform a t i ~ nanalogous , ~ ~ ~ to the situation for structures I and la. However, with respect to the lower energy reference structure corresponding to the completely trans-staggered conformation of propane-l,3-di01,~~” a small 0 ” H bond weakening is im-

plied. Irrespective of the sign, the magnitude of the change in OH-bond population is seen to be relatively small for all of the intramolecular cases. The intramolecular “hydrogen bond” energies listed in Table IV, based on the reference structures defined above, clearly do not make possible a direct comparison of the intrinsic strengths of inter- and intramolecular hydrogen bonding. The dependence of the intramolecular “hydrogen-bond’’ energy on the reference structure is particularly emphasized by the case of propane- 1,3-diol. These energy differences are, of course, composites of various short- and long-range factors, as noted above. Rotating $ 1 from 0 to 180’ in glycolic acid (I Ja) in addition to facilitating the 0 ” H - 0 interaction might be expected to reduce the nonbonded interactions between 0” and 0 (the lone pairs of 0” are directed away from 0 when $1 = 0’) and lead to a more favorable dipole-dipole interaction with the carbonyl bond (see discussion of I11 below); however, this process is also seen to entail eclipsing of the 0 ” H and C C bonds. The net result is an energy lowering of 6.1 kcal/mol. With regard to the O”-O interaction, the overlap population gives no indication of significant repulsion in spite of the short separation (2.73 A). In fact, the population is slightly positive (0.010) for $ 1 = 0 (cf., -0.012 for 41 = 180’). Overlap populations are, of course, not definitive indications of interaction energies. The only suggestion of O”.-O repulsion is the slight increase in the 0 ” C C bond angle exhibited by many of the molecules under discussion.’-I0 Typical values are -1 12’ (vs. a standard value of -1 However, this increase does not appear to be correlated with the $ 1 angle, and a few examples of unusually small 0”CC angles (-107-108’) are also found in the tartaric acids3 and the citrates9 C. Further Comparisons of Conformational Energetics. Glycolaldehyde (111) offers interesting similarities and contrasts in conformational behavior, relative to that of glycolic acid. It is somewhat simpler to analyze from a theoretical point of view, primarily since it involves the interaction of only a single pair of CO bonds. It also has the advantage of having been studied experimentally in the gas phase,l2Sl3 thus providing data for direct comparison with calculations on the isolated species. Not surprisingly, the relative conformational energies of glycolic acid and glycolaldehyde are quite similar for small values of 42 (Figures 3 and 4),44with glycolaldehyde exhibiting the same (but somewhat less extended) elongated basin (cf. Figures 2 and 3), centered about the minimumenergy structure 111, whose weak hydrogen bonding has already been discussed in Section VB (see Table IV). Glycolaldehyde possesses a second local minimum a t $ 2 = 180’ (2.3 kcal/mol above the $2 = 0’ isomer), but here it differs sharply from glycolic acid, since for 42 = 180’ it prefers a trans-staggered O”H conformation = 0’) in the absence of the possibility of H bonding. This divergence of behavior at large 42 is illustrated in terms of the crossing of the two onedimensional profiles (with $1 fixed a t 180’) as 42 increases in Figure 5a (the “minimum energy” paths by definition lead to the proper local minima a t $2 = 180’ ). The second isomer of glycolaldehyde has not been observed so far in the gas phase despite a careful search in the microwave experiments.l0 The corresponding pair of isomers for the isoelectronic a-aminoacetaldehyde is calculated (using standard bond lengths and angles)30 to have a separation of -I .4 kcal/mo1,26 a remarkable result in light of the fact that all three bonds a t the nitrogen are eclipsed with other single bonds ( C C and C H ) in the higher energy 41 = 0, 42 = 180’ isomer.45 The most direct comparison between the a-halo carbonyl systems and glycolic acid and glycolaldehyde is revealed by the nonhydrogen-bonded (i.e., $1 = 0’) portions of the potential energy surfaces displayed in Figure 5b. (The sensitivity of the conformational energy with respect to $2 is emphasized by comparison with the analogous profile for unsubstituted ac-

-

Newton, Jeffrey

/ Stereochemistry of a-Hydroxycarboxylic Acids

2420

( 0 )

Iplt0+= 3.OD

H

P

(c)

I p l to+ = 3 . 4 0

Figure 6. Schematic representation of dipole-dipole interactions for selected conformations of glycolaldehyde: (a) $ 1 = 180, $2 = 0"; (b) $ 1 = 0, $ 2 = 0"; (c) $ 1 = 0, $ 2 = 180'. The I@ltut values refer to the magnitude of the calculated (4-3 IC) dipole moments. The local aldehyde and alcohol dipole moments are 3.3 and 2.1 D,respectively. as discussed in the text.

etaldehyde.) Relative to 42 = O", glycolic acid is seen to have a second minimum a t slightly higher energy (0.7 kcal/ mol) located a t 1 50°, the same angle observed for the second isomer of the nonfluoro a-haloacetyl halides.23 The small barrier (-0.3 kcal/mol) associated with the 42 = 180" conformation is also qualitatively consistent with the microwave analysis of the fluoro d e r i ~ a t i v e In . ~ ~the case of glycolaldehyde, however, the constraint of 4)= 0" causes the trans isomer (42 = 180') to be -4.6 kcal/mol lower than the cis isomer (42 = OO), in quantitative agreement with corresponding calculations for fluoroacetaldehyde (-4.1 kcal/ mo1)I6 and in qualitative agreement with the known energetics of rotational isomers in a-halo ketones (see Section iI).47 Thus in the case of glycolic acid, the possibility of hydrogen bonding (41 = 180') reinforces the preference for 42 = 0" over 42 = 150- 180" exhibited by the 41= 0" conformations and by the a-haloacetyl halides and related systems. On the other hand, H bonding in glycolaldehyde (for 42 = 0") reverses the energy ordering for 42 = 0" and 42 = 180" displayed by the 41 = 0" isomers and by the a-halo aldehydes and ketones. An important factor in the preference for 4 2 = 180" over 4 2 = 0" in glycolaldehyde (with 41 fixed at 0") is the interaction between the dipole moments of the aldehyde ( p = 3.3 D) and alcohol ( p = 2.1 D) moieties (see Figure 6);48they are nearly parallel in the completely trans-staggered conformation (41= 42 = O"), with a total dipole moment of 5.3 D, and roughly perpendicular for the 41 = 0,42 = 180" (total dipole = 3.4 D).49 The local bond moments are also nearly perpendicular in the minimum energy, H-bonded structure 111, whose calculated dipole moment is 3.0 D, compared with the macrowave value of 2.34 D. The calculated dipole moments for acetaldehyde and ethanol are exaggerated by a similar amount (-25%) as is typical for the 4-31G or other split valence basis et^.^^^,^,^^ Journal of the American Chemical Society

/ 99:8 /

VI. Summary Ab initio calculation of the conformational potential energy surfaces for various a-hydroxycarbonyl systems has led to the following conclusions: (1) The minimum-energy conformation of the isolated species corresponds to an internally hydrogenbonded geometry (41 = 180" and 42 = 0"). (2) Departures from this conformation in the 41direction generally cost less energy than do similar variations of 42, a result consistent with the experimental crystal data for the neutral acids arid carboxylate anions, where the formation of intermolecular hydrogen bonding is seen to entail conformational changes in the $1 coordinate, with the 42 values remaining clustered about zero. (3) Although the relative stabilities associated with structures la and 111 may be described partly in terms of internal hydrogen bonding, analysis of the charge distribution indicates that the short-range (0"H-0) effects are smaller than those generally found in intermolecular H bonding. In particular, no weakening of the O"H bond is found when la and 111 are compared with non-H-bonded reference conformers. (4) The isolated glycolic acid and glycolaldehyde molecules have local conformational minima for 42 close to 180" within -2--3 kcal/mol of the lowest energy 42 = 0" conformations, and hence they should be amenable to experimental detection. (5) It is emphasized that all low-lying local minima in both molecules correspond to a planar framework ($2 = 0 or 180") even in the absence of hydrogen bonding, a situation which most likely indicates the importance of dipolar interactions in the conformational energetics. (6) Selected sections of the potential energy surfaces for 1 and I I I are consistent with known energetics of rotational isomers in related systems where H bonding is not possible. Acknowledgment. Research was performed under the auspices of the U S . Energy Research and Development Administration. W e are grateful to Professor J . A. Pople and Dr. S. Vishveshwara for numerous illuminating discussions and iriformation prior to publication, and to Professor P. A. Kollman for correspondence and a preprint of ref 42b. References and Notes (1) (a)G.A.JeffreyandG.S.Parry,Nature(London), 169, 1105(1952).(b)More recently, a brief discussion of this and related conformational phenomena has been given by J. A. Kanters. J. Kroon, A. F. Peerdeman, and J. C. Schoone. Tetrahedron, 23, 4027 (1967). (2) R. D. Ellison, C. K. Johnson, and H. A. Levy, Acta Crystaf/ogr., Sect. B. 27, 333 (197 1). (3) (a) Y. Okaya. N. R. Stemple, and M. I.Kay, Acta Crystaffogr., 21,237 (1966); (b) G. A. Bootsma and J. C. Schoone. ibid., 22, 522 (1967). (4) 8. P. Van Eijck, J. A. Kanters. and J. Kroon. Acta Crystalfogr., 19, 435 (1965). (5) (a) J. P. Glusker, J. A. Minkin, and A. L. Patterson, Acta Crystaflogr., Sect. -3, 25, 1066 (1969); (b) G. Roelofsen and J. A. Kanters, Cryst. Struct. Commun., 1, 23 (1972). (6) R. H. Colton and D. E. Herr, Acta Crystaffogr., 18, 820 (1965). (7) (a) G.K. Ambady and G. Kartha, Acta Crystalfog., Sect.6, 24, 1540 (1968); (b) A. J. van Bommel and J. M. Bijvoet, Acta Crystallogr., 11, 6 1 (1958); (c) J. Kroon, A. F. Peerdeman. and J. M. Bijvoet, ibid., 19,293, (1965); (d) J. Kroon and J. A. Kanters, Acta Crystallogr., Sect. 8, 28, 714 (1972): M. Currie, J. C. Speakman, J. A. Kanters, and J. Kroon, J. Chem. Soc., Perkin Trans. 2, 1549 (1975). (6) N. C. Panagiotopoulos, G. A. Jeffrey, S. J. La Placa, and W. C. Hamilton, Acta Crystallogr., Sect. B, 30, 1421 (1974). (9) (a) J. P. Glusker, D. van der Helm, W. E. Love, M. L. Dornberg. J. A. Minkin, C. K. Johnson, and A. L. Patterson, Acta Crystaffogr., 19, 561 (1965): (b) E. J. Gabe, J. P. Glusker, J. A. Minikin, and A. L. Patterson, ibid., 22, 366 (1967). (10) (a) K.-M. Marstokk and H. Mgllendal. J. Mol. Struct., 5 , 205 (1970); (b) K. M. Marstokk and H. Mgllendal, ibid., 16, 259 (1973). (1 1) H. Michelsen and P. Klaboe, J. Mol. Struct., 4, 293 (1969). (12) Zero values of $, ( i = 1, 2, 3) correspond to staggered conformations for the single bond sequences HO"CC, O"CCO', and CCO'H, respectively. (In the case of the carboxylate anions, the CO' bond has partial double bond character due to resonance.) The values of 4, increase in the sense implied by the arrows in structure I; the arrows refer to rotation of the 0"H. CO'. and O'H bonds about the 0°C. CC, and CO' bonds, respectively. As defined here, the 6, are the negative supplements of the associated torsion angles, whose zero values correspond to eclipsed conformations. The reader should note that due to small deviations from local planarity at the carbonyl carbon in the crystal structures, a value of $2 = 0" does not necessarily

April 13, 1977

242 1 imply perfect eclipsing of the CO" and CO bonds. (13) Q. C. Pimentel and A. L. McClellan, "The Hydrogen Bond", W. H. Freeman, San Francisco, Calif., 1960. (14) L.,Radom,W. A. Lathan, W. J. Hehre, and J. A. Pople, Aust. J. Chem., 25, 1601 (1972). A more detailed study of the conformational energy of glycolaldehyde is presented here. Note that a slightly different framework geometry is employed in the present work, as discussed below. [ I s ) (a),R. Ditchfield, W. J. Hehre, and J. A. Pople, J. Chem. Phys., 54, 724 (1971). (b) The 4-31G level has been,extensivelydocumented as a reliable tool for studying conformational energiesl4.l6and hydrogen-bonded int e r a c t i o n ~ .(c) ' ~ Preliminary calculations on glycolic acid were carried out with t k smaller STO-3G basis (W. J. Hehre, R. F. Stewart, and J. A. Pople, ibid., 51, 2657 (1969)).While most of the results were in qualitative accord with those found at the 4-31G level, significant quantitative differences were observed, and we limit our discussion to the results based on the more flexible 4-31G basis. (16) (a) L. Radom, W. A. Lathan, W. J. Hehre, and J. A. Pople, J. Am. Chem. SOC., 95,693 (1973); (b) ibid., 95,699 (1973); (c) J . A. Pople, Tetrahedron, 30, 1605 (1974). (17) (a) J. E. Del Bene and J. A. Pople, J. Chem. Phys. 58, 3605 (1973); (b) J. E. Del Bene, Chem. Phys. Lett., 24, 209 (1974); (c) W. C. Topp and L. C. Allen, J. Am. Chem. Soc., 96, 5291 (1974); (d) P. Kollman, J. McKelvey, A. Johansson. and S. Rothenberg, ibid., 97, 955 (1975); (e) J. D. Dill, L. C. Allen, W. C. Topp, and J. A. Pople, ibid., 97, 7220 (1975). (18) R. W. Kilb, C. C. Lin, and E. B. Wilson Jr., J. Chem. Phys., 26, 1695 (1957). (19) (a) W. J. Hehre and L. Salem, J. Chem. SOC.,Chem. Commun., 754 (1973); (b) W. J. Hehre. J. A. Pople, and A. J. P. Devaquet, J. Am. Chem. SOC.,98, 664 (1976). (20) R. B. Davidson and L. C. Allen, J. Chem. Phys., 54, 2828 (1971). (21) (a) S.4. Mizushima,T. Shimanouchi, T. Miyazawa, I. Ichishima, K. Kuratani, I. Nakagawa, and N. Shido, J. Chem. Phys., 21, 815 (1953); (b) G. A. Crowder and B. R. Cook, ibid., 46, 367 (1967); (c) E. Saegebarth and L. C. Krisher, ibid., 52, 3555 (1970). (22) R. G. Ford, J. Chem. Phys., 65, 354 (1976). (23) I. Naka,gawa, I. Ichishima, K. Kuratani, T. Miyazawa, T. Shimanouchi, and S.4. Mizushima, J. Chem. Phys., 20, 1720 (1952). (24) E. Saegebarth and E. R. Wilson Jr., J. Chem. Phys., 46, 3088 (1967). (25) B. P. Van Eijck, G. Van Der Plaats, and P. H. V2n Roon, J. Mol. Struct., 11, 67 (1972). (26) S. Vishveshwara and J. A. Pople, J. Am. Chem. Soc.. following paper in this issue. (27) (a) One example of 43 close to 180' can be inferred from the data of ref 7d, but this involves a strong ionic intermolecular interaction (see footnote fof Table I)and accordingly has not been included in Table I. (b) The occurrence of 43 = 180' isomers in the related species, cr-alkoxyacetic acids, has been discussed by M. Oki and M. Hirota, Bull. Chem. SOC.Jpn., 36, 290 (1963). See also ref 25. (28) (a),L. Pauling, "The Nature of the Chemical Bond', Cornell University Press, Ithaci, N.Y., 1960, p 257 ff; (b) I. D. Brown, Acta Crystaliogr., Sect. A, 32, 24 (1976), has recently given a detailed dlscussion of 0-0 contacts in QH-0 hydrogen bonds. (29) G. A. Jeffrey, Carbohydr. Res., 28, 233 (1973). (30) J. A. Pople and M. S. Gordon, J. Am. Chem. SOC., 89,4253 (1967). (31) E. V. lvash and D. M. Dennison, J. Chem. Phys., 21, 1804(1953). (32) There is also some indication of a small red shift in the OH stretching frequency (a standard characteristic of hydrogen bonding),l 3 as inferred by comparison with ethanol,'O but no intramolecular red shift data based on other conformations is available. (33) See summary and discussion of data given by A. F. Beecham, A. C. Hurley, M. L. Mackay, V. W. Maslen, and A. McL: Matheson, J. Chem. Phys., 49, 3312 (1968). (34) (a) P. A. Kollman and L. C. Allen, J. Am. Chem. Soc., 92,6101 (1970); (b) M.D. Newton and S. Ehrenson, ibid., 93,4d71 (1971). (35) P. A. Kollman and L. C. Allen, J. A h . Chem. SOC.,92, 753 (1970). (36) (a) R. E. Penn and R. F. Curl Jr., J. Chem. Phys., 55, 651 (1971); (b) R. G. Azrak and E. B. Wilson, ibid., 52,5299 (1970); (c) K . S. Buckton and R. G. Azrak, ibid., 52, 5652 (1970). (d) A note of caution regarding the postulation of internal H b ~ n d i n g ~ is ~= implied - ~ by the results of R. C. Griffith and J. D. Roberts, Tetrahedron Lett., 39, 3499 (1974). Their proton NMR work on dilute solutions of 2-fluoroethanol in carbon tetrachloride gave no indication of significant internal hydrogen bonding (the proton chemical shift was very similar to that of ethanol) in spite of a small red shift in the OH stretching frequency (relative to that for the trans conformation) detected

in infrared solution Studies (P. J. Krueger and H. D. Mettee, Anal. J. Chem., 42, 326 (1964)); this latter paper draws attention to the fact that OH frequency red shifts may not be a reliable indicator of H-bond strength, in cases of very weak intramolecular interactions. (37) Since the 4-31G basis is expected to exagierate the equilibrium HOC bond angle (the 4-31G methanol value is -1 13 , compared to an accurate microwave value bf 10603'), we note that even when the glycolaldehyde HO"C angle was held at the reported microwave value of 102°,'0a the calculations still gave no indication of appreciable O"H stretching (a value of 0.96 A was obtained]. Furthermore, the calculated O"H and +HO"C values were virtually insensitive to the small differences between the experimental valuesioa and the values adopted here for the other structural parameters of the molecule. (38) Note that the potential energy surfaces displayed in Figures 2 and 3 are = 180°, $2 = invariant with respect to inversion through the center (41 0'). This invaliance is not a property of the experimental data due to the presence of the crystalline environment. (39) For a discussion of angles associated with hydrogen-bonded carboxyl oxygens see ref 17b, 17d, K. Morokuma, J. Chem. Phys., 55, 1236 (1971), and J. E. Del Bene, ibid., 62, 1314 (1975). The strained angles quoted for structure la are quite similar to those found experimentally'0a for glycolaldehyde (structure ill). (40) A. Pullman and B. Pullman, Q. Rev. Biophys., 7, 505 (1975). (41) We wish to acknowledge the reteree for noting this point and for making several other helpful suggestions. (42) (a) For example, propane-1,3-diol has been studied by A. Johansson, P. A. Kollman, and S. Rothenberg, Chem. Phys. Leff., 18,276 (1973).We have carried out calculations for this molecule at the 4-31G level using somewhat different values of the conformational angles. The H-bonded structure employed gauche (60') conformations for the two OCCC moieties and for the HOCC proton donor end of the molecule: the other HOCC group was trans staggered and the other angles are as given by Kollman et al.; the 4CCC was 109.47'. Two non-H-bonded reference structures were considered: (i) the same carbon-oxygen framework as above, but with both HOCC groups trans staggered; and (ii) the completely trans-staggered conformation; the latter one is more stable, as indicated in Table IV. It must be emphasized that the results for propane-l,3dioI are used here only for illustrative ,purposes. A detailed examination of the conformational surface would undoubtedly reveal a departure from perfect gauche conformations due to the oxygen-oxygen conformations. Kollman et al. have also pointed out that bond angles might be expected to open up for similar reasons. (b) A study of substituted phenols has recently been completed by S. W. Dietrich, E. C. Jorgensen, P. A. Kollman, and S.Rothenberg, submitted for publication. (c) The possibility of stronger, symmetrical internal hydrogen bonding in $-dicarbonyls has led to recent studies by G. Karlstrom, H. Wennerstrom. B. hnsson, S. Forsen, J. Almlof, and B. Roos, J. Am. Chem. SOC.,97,4188 (1975),and A. D. lsaacson and K. Morokuma, ibid., 97,4453 (1975). (43) G. A. Jeffrey and J. S. Kim, Carbohydr. Res., 14, 207 (1970); 15, 310 (1970). (44) The energy variation with respect to is compared with the corresponding behavior for a simple alcohol (CzH50H)in Figure 4. (45) = 0' is defined analogously to the case of glycolic acid (I) and glycolaldehyde (Ill),l 2 except that the bisector of the HNH angle of the amino group takes the place of the O"H bond. (46) The microwave studyz4of fluoroacetal fluoride suggested a local minimum at $2 = 180' but could not rule out a small bump centered at $2 = 180'. An increase in $2 (relative to 150') in going from OH and the larger halogens to fluorine may well reflect the somewhat smaller van der Waals rad i d 8 of the latter substitutent. (47) A smaller energy separation for the ketones relative to the aldehydes may be expected, since the ketone methyl group leads to greater nonbonded repulsion with the a-substituent than does the aldehyde proton in the (bZ = 180' isomer. (48) This effect more than offsets the eclipsing of the CO single bonds in the I $ ~ = 180' conformation. (49) The magnitude and direction of the local bond moments are based on 4-31G calculations for acetaldehyde and ethanol. The geometrical parameters were the same as those used in the glycolaldehyde calculation,as described in Section IV. (50) Although inclusion of polarization functions generally improves the 4-3 1G dipole moments for simple hydrides,"' 6-31G' calculations for glycolaldehyde have a rather small effect on the dipole moments (relative to the 4-31G results).26

Newton, Jeffrey

/ Stereochemistry of a-Hydroxycarboxylic Acids